Re: FF layer restrictions [Was: [PATCH 1/1] Input: add sensable phantom driver]

From: Jiri Slaby <jirislaby@gmail.com>
To: "\"STenyaK (Bruno González)\"" <stenyak@gmail.com>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>,
	Anssi Hannula <anssi.hannula@gmail.com>,
	johann deneux <johann.deneux@gmail.com>,
	Linux kernel mailing list <linux-kernel@vger.kernel.org>,
	linux-input@atrey.karlin.mff.cuni.cz
Subject: Re: FF layer restrictions [Was: [PATCH 1/1] Input: add sensable phantom driver]
Date: Wed, 21 Mar 2007 14:31:08 +0100	[thread overview]
Message-ID: <4601339C.6010706@gmail.com> (raw)
In-Reply-To: <8e4ff20a0703141218r3add7923n74470b5e16f58c5c@mail.gmail.com>

STenyaK (Bruno González) napsal(a):
> On 3/14/07, Dmitry Torokhov <dmitry.torokhov@gmail.com> wrote:
>> > I have a question: if the force is to be 3D, why only 3 possible
>> values?
>> > What would they be, 3 torques or 3 forces? In the case of car sims (ff
>> > steering wheels), only one axis of torque is usually used (except
>> for 6 dof
>> > platforms, as mentioned).
>> >
>>
>> I wonder if we could somehow extend or augment FF envelope se we could
>> specify a plane for the effect.. Then a vector could be represented by
>> a sum 3 constant effects in 3 separate planes and we could also use
>> spring and other effects as well.
> 
> Ideally, afaik we should use:
> -3 values for translation force (linear force): x,y,z components of
> the force vector.
> -4 values for rotation force (torque): x,y,z,w components of the
> quaternion. You can also use euler angles (and i think there are
> another one or two notations), which is just 3 values, but i'm not
> sure it will be a correct decision (due to the gimbal lock problem,
> which may or may not be present in ff devices, dunno).

So, the resolution is? Since I want no longer have out-of-kernel driver, I
need to know how to implement the phantom driver -- merge it `as is', i.e.
control through mmio or rewrite ff layer somehow, and in that case how?

There seem to be only few possibilities:

- new 3D effect, which will be problematic in any other future use, that may
need more than 3 axis. no matter if torques or vector is passed -- depending
on device and programmer (as I need to compute torques from forces in FP).
Maybe struct with 2x 3 axis is OK

- "raw" effect, which may contain more axis, but this is ugly in Anssi's eyes

- something else? (did I forget something)

thanks,
-- 
http://www.fi.muni.cz/~xslaby/            Jiri Slaby
faculty of informatics, masaryk university, brno, cz
e-mail: jirislaby gmail com, gpg pubkey fingerprint:
B674 9967 0407 CE62 ACC8  22A0 32CC 55C3 39D4 7A7E